# Vext Audit

> Comprehensive security audit of your OpenClaw installation — checks configuration, file permissions, network exposure, SOUL.md integrity, and installed skill safety. Built by Vext Labs.

- Skill: `dvcrn/vext-audit` (Agent Skill, multi-file: 2 files)
- Install (CLI): `npx skillmds@latest add dvcrn/vext-audit`
- Raw SKILL.md: https://api.skillmd.com/api/skills/dvcrn/vext-audit/raw
- Safety review: pending (external: skill-scanner PASS, skillspector PASS)
- Works with: Claude Code, Claude.ai, OpenAI Codex
- Category: Security
- Author: dvcrn (https://skillmd.com/u/dvcrn)
- Updated: 2026-09-08
- Page: https://skillmd.com/skills/dvcrn/vext-audit

---


# VEXT Audit

Run a full security audit of your OpenClaw installation. Checks configuration best practices, file permissions, network exposure, identity file integrity, and scans all installed skills.

## Usage

- "Audit my OpenClaw installation"
- "Run a security audit"
- "Check my OpenClaw security posture"

## What It Checks

1. **Configuration** — sandbox mode, tool permissions, network egress, API key storage, binding address
2. **File permissions** — ensures sensitive files aren't world-readable/writable
3. **Network exposure** — checks for 0.0.0.0 binding and exposed ports
4. **SOUL.md integrity** — scans for injected prompt injection patterns
5. **Installed skills** — runs the full VEXT Shield scanner on all skills
6. **Security grade** — computes an A-F letter grade with score

## Rules

- Only perform read-only checks — never modify configuration or files
- Report all findings honestly
- Do not transmit audit results externally
- Always provide actionable remediation steps

## Safety

- This skill only reads files and configuration — no modifications
- No network requests are made
- Audit reports are saved locally to ~/.openclaw/vext-shield/reports/